Resting state functional magnetic resonance imaging (rs-fMRI) contains a wealth of information about the large-scale structure of neural activity in the brain, an area that has been relatively unexplored. Rs-fMRI has provided some insight into the macroscopic organization of brain activity by identifying functional networks that are reproducible across subjects. The functional networks are often interpreted as if they represent time- varying interactions between areas of the type that would be expected to arise from cognitive processes, but the same network structure can be found in conditions where cognition is suppressed or absent (sleep, coma, and anesthesia). This persistent network structure is one of the lingering puzzles in rs-fMRI. Our previous work has shown that large-scale spatiotemporal quasi-periodic patterns (QPPs) of electrical activity can be isolated from the BOLD signal, allowing us to separate slow, semi-periodic modulations from the more localized aperiodic activity that is expected to arise from cognition and information processing. This led us to hypothesize that the QPPs account for a persistent background pattern of neuromodulation, over which time-varying contributions from cognition and information processing are superimposed. Our preliminary data indicates that QPPs arise from a different type of brain activity than the neural activity linked to information processing and cognition but still account for a substantial portion of the functional connectivity in the brain. In Aim 1 , we extend our previous work to investigate the neurophysiological sources that play a role in QPP generation using multimodal imaging in the rat. Our working model is that the QPPs arise from localized input from subcortical nuclei that then propagates across the cortex through the coordinated actions of neurons and astrocytes. In humans, the QPPs are most dominant in the default mode network (DMN), a critical structure implicated in numerous functions and altered in many disorders. Our preliminary data shows that QPPs account for a substantial portion of the connectivity in the DMN. In Aim 2, we will compare functional network metrics throughout the brain before and after the QPPs are removed by regression to determine how the presence of the infraslow modulation impacts standard analysis. Our final aim directly examines the hypothesis that QPPs account for background activity over which time- varying activity more relevant to cognition is superimposed. We will calculate the relative contribution of QPPs to the BOLD signal as a function of anesthetic depth in rats, where we expect their contribution to increase as anesthetic depth increases, and during tasks with varying difficulty in humans, where we expect their relative contribution to decrease as a function of increasing cognitive demand. Taken together, the work in this proposal will change the way we interpret rs-fMRI by allowing separate examination of two distinct components of brain activity that may both be of clinical interest.